UFC Data Visualisation

library(tidyverse)

## ── Attaching packages ────────────────────────────────────────────────────────── tidyverse 1.2.1.9000 ──

## ✔ ggplot2 3.2.1     ✔ purrr   0.3.3
## ✔ tibble  2.1.3     ✔ dplyr   0.8.3
## ✔ tidyr   1.0.0     ✔ stringr 1.4.0
## ✔ readr   1.3.1     ✔ forcats 0.4.0

## ── Conflicts ────────────────────────────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag()    masks stats::lag()

library(dplyr)
library(janitor)

## 
## Attaching package: 'janitor'

## The following objects are masked from 'package:stats':
## 
##     chisq.test, fisher.test

library(fastDummies)
library(GGally)

## Registered S3 method overwritten by 'GGally':
##   method from   
##   +.gg   ggplot2

## 
## Attaching package: 'GGally'

## The following object is masked from 'package:dplyr':
## 
##     nasa

library(lubridate)

## 
## Attaching package: 'lubridate'

## The following object is masked from 'package:base':
## 
##     date

library(ggradar)
require(Hmisc)

## Loading required package: Hmisc

## Loading required package: lattice

## Loading required package: survival

## Loading required package: Formula

## 
## Attaching package: 'Hmisc'

## The following objects are masked from 'package:dplyr':
## 
##     src, summarize

## The following objects are masked from 'package:base':
## 
##     format.pval, units

require(digest)

## Loading required package: digest

require(cluster)    

## Loading required package: cluster

require(factoextra) 

## Loading required package: factoextra

## Welcome! Related Books: `Practical Guide To Cluster Analysis in R` at https://goo.gl/13EFCZ

library(gridExtra)

## 
## Attaching package: 'gridExtra'

## The following object is masked from 'package:dplyr':
## 
##     combine

library(fmsb)
library(ggiraph)
library(moonBook)

## 
## Attaching package: 'moonBook'

## The following object is masked from 'package:lattice':
## 
##     densityplot

library(reshape2)

## 
## Attaching package: 'reshape2'

## The following object is masked from 'package:tidyr':
## 
##     smiths

library(sjmisc)

## 
## Attaching package: 'sjmisc'

## The following object is masked from 'package:Hmisc':
## 
##     %nin%

## The following objects are masked from 'package:janitor':
## 
##     remove_empty_cols, remove_empty_rows

## The following object is masked from 'package:purrr':
## 
##     is_empty

## The following object is masked from 'package:tidyr':
## 
##     replace_na

## The following object is masked from 'package:tibble':
## 
##     add_case

library(ggiraphExtra)

## 
## Attaching package: 'ggiraphExtra'

## The following objects are masked from 'package:moonBook':
## 
##     addLabelDf, getMapping

library(scales)

## 
## Attaching package: 'scales'

## The following object is masked from 'package:moonBook':
## 
##     comma

## The following object is masked from 'package:purrr':
## 
##     discard

## The following object is masked from 'package:readr':
## 
##     col_factor

library(extrafont)

## Registering fonts with R

library(ggplot2)

1 Introduction: The Story Behind UFC

2 Data Cleaning and Processing

data <- read_csv("data.csv")

## Parsed with column specification:
## cols(
##   .default = col_double(),
##   R_fighter = col_character(),
##   B_fighter = col_character(),
##   Referee = col_character(),
##   date = col_date(format = ""),
##   location = col_character(),
##   Winner = col_character(),
##   title_bout = col_logical(),
##   weight_class = col_character(),
##   B_Stance = col_character(),
##   R_Stance = col_character()
## )

## See spec(...) for full column specifications.

clean_names(data) 

## # A tibble: 5,144 x 145
##    r_fighter b_fighter referee date       location winner title_bout
##    <chr>     <chr>     <chr>   <date>     <chr>    <chr>  <lgl>     
##  1 Henry Ce… Marlon M… Marc G… 2019-06-08 Chicago… Red    TRUE      
##  2 Valentin… Jessica … Robert… 2019-06-08 Chicago… Red    TRUE      
##  3 Tony Fer… Donald C… Dan Mi… 2019-06-08 Chicago… Red    FALSE     
##  4 Jimmie R… Petr Yan  Kevin … 2019-06-08 Chicago… Blue   FALSE     
##  5 Tai Tuiv… Blagoy I… Dan Mi… 2019-06-08 Chicago… Blue   FALSE     
##  6 Tatiana … Nina Ans… Robert… 2019-06-08 Chicago… Red    FALSE     
##  7 Aljamain… Pedro Mu… Marc G… 2019-06-08 Chicago… Red    FALSE     
##  8 Karolina… Alexa Gr… Kevin … 2019-06-08 Chicago… Blue   FALSE     
##  9 Ricardo … Calvin K… Dan Mi… 2019-06-08 Chicago… Blue   FALSE     
## 10 Yan Xiao… Angela H… Robert… 2019-06-08 Chicago… Red    FALSE     
## # … with 5,134 more rows, and 138 more variables: weight_class <chr>,
## #   no_of_rounds <dbl>, b_current_lose_streak <dbl>,
## #   b_current_win_streak <dbl>, b_draw <dbl>, b_avg_body_att <dbl>,
## #   b_avg_body_landed <dbl>, b_avg_clinch_att <dbl>, b_avg_clinch_landed <dbl>,
## #   b_avg_distance_att <dbl>, b_avg_distance_landed <dbl>,
## #   b_avg_ground_att <dbl>, b_avg_ground_landed <dbl>, b_avg_head_att <dbl>,
## #   b_avg_head_landed <dbl>, b_avg_kd <dbl>, b_avg_leg_att <dbl>,
## #   b_avg_leg_landed <dbl>, b_avg_pass <dbl>, b_avg_rev <dbl>,
## #   b_avg_sig_str_att <dbl>, b_avg_sig_str_landed <dbl>,
## #   b_avg_sig_str_pct <dbl>, b_avg_sub_att <dbl>, b_avg_td_att <dbl>,
## #   b_avg_td_landed <dbl>, b_avg_td_pct <dbl>, b_avg_total_str_att <dbl>,
## #   b_avg_total_str_landed <dbl>, b_longest_win_streak <dbl>, b_losses <dbl>,
## #   b_avg_opp_body_att <dbl>, b_avg_opp_body_landed <dbl>,
## #   b_avg_opp_clinch_att <dbl>, b_avg_opp_clinch_landed <dbl>,
## #   b_avg_opp_distance_att <dbl>, b_avg_opp_distance_landed <dbl>,
## #   b_avg_opp_ground_att <dbl>, b_avg_opp_ground_landed <dbl>,
## #   b_avg_opp_head_att <dbl>, b_avg_opp_head_landed <dbl>, b_avg_opp_kd <dbl>,
## #   b_avg_opp_leg_att <dbl>, b_avg_opp_leg_landed <dbl>, b_avg_opp_pass <dbl>,
## #   b_avg_opp_rev <dbl>, b_avg_opp_sig_str_att <dbl>,
## #   b_avg_opp_sig_str_landed <dbl>, b_avg_opp_sig_str_pct <dbl>,
## #   b_avg_opp_sub_att <dbl>, b_avg_opp_td_att <dbl>, b_avg_opp_td_landed <dbl>,
## #   b_avg_opp_td_pct <dbl>, b_avg_opp_total_str_att <dbl>,
## #   b_avg_opp_total_str_landed <dbl>, b_total_rounds_fought <dbl>,
## #   b_total_time_fought_seconds <dbl>, b_total_title_bouts <dbl>,
## #   b_win_by_decision_majority <dbl>, b_win_by_decision_split <dbl>,
## #   b_win_by_decision_unanimous <dbl>, b_win_by_ko_tko <dbl>,
## #   b_win_by_submission <dbl>, b_win_by_tko_doctor_stoppage <dbl>,
## #   b_wins <dbl>, b_stance <chr>, b_height_cms <dbl>, b_reach_cms <dbl>,
## #   b_weight_lbs <dbl>, r_current_lose_streak <dbl>,
## #   r_current_win_streak <dbl>, r_draw <dbl>, r_avg_body_att <dbl>,
## #   r_avg_body_landed <dbl>, r_avg_clinch_att <dbl>, r_avg_clinch_landed <dbl>,
## #   r_avg_distance_att <dbl>, r_avg_distance_landed <dbl>,
## #   r_avg_ground_att <dbl>, r_avg_ground_landed <dbl>, r_avg_head_att <dbl>,
## #   r_avg_head_landed <dbl>, r_avg_kd <dbl>, r_avg_leg_att <dbl>,
## #   r_avg_leg_landed <dbl>, r_avg_pass <dbl>, r_avg_rev <dbl>,
## #   r_avg_sig_str_att <dbl>, r_avg_sig_str_landed <dbl>,
## #   r_avg_sig_str_pct <dbl>, r_avg_sub_att <dbl>, r_avg_td_att <dbl>,
## #   r_avg_td_landed <dbl>, r_avg_td_pct <dbl>, r_avg_total_str_att <dbl>,
## #   r_avg_total_str_landed <dbl>, r_longest_win_streak <dbl>, r_losses <dbl>,
## #   r_avg_opp_body_att <dbl>, r_avg_opp_body_landed <dbl>, …

data <- data %>% 
  mutate(women = ifelse(grepl("Women", weight_class), TRUE, FALSE))

fight_data <- data %>% 
  dplyr::select(1:9) 

#Splitting into men and Women
data_men <- data %>% 
  subset(women == FALSE)
data_women <- data %>% 
  subset(women == TRUE)

#Splitting data into Fight and Fighter Data
blue_profile_data <- data_men %>% 
  dplyr::select(Referee, date, location, starts_with("B_")) %>% 
  mutate(color = "blue")
  
red_profile_data <- data_men %>% 
  dplyr::select(Referee, date, location, starts_with("R_")) %>% 
  mutate(color = "red")

#Cleaning column names
names(blue_profile_data) <- ifelse(grepl("B_", names(blue_profile_data)), substring(names(blue_profile_data), 3), names(blue_profile_data))

names(red_profile_data) <- ifelse(grepl("R_", names(red_profile_data)), substring(names(red_profile_data), 3), names(red_profile_data))

#Unifying the Fighter Datasets
fighter_profile_data <- rbind(blue_profile_data, red_profile_data) %>%
  mutate(Stance = as.factor(Stance)) %>% 
  rename(win_by_ko_tko = `win_by_KO/TKO`,
         total_time_fought_seconds = `total_time_fought(seconds)`)

# Turning data relative to number of fights
fighter_profile_data <- fighter_profile_data %>% 
  mutate(win_by_Decision_Majority = win_by_Decision_Majority/wins,
         win_by_Decision_Split = win_by_Decision_Split/wins,
         win_by_Decision_Unanimous = win_by_Decision_Unanimous/wins,
         win_by_Submission = win_by_Submission/wins,
         win_by_ko_tko = win_by_ko_tko/wins,
         win_by_TKO_Doctor_Stoppage = win_by_TKO_Doctor_Stoppage/wins
         ) %>% 
  mutate(win_by_ko_tko = win_by_ko_tko + win_by_TKO_Doctor_Stoppage,
         win_by_Decision = win_by_Decision_Majority + win_by_Decision_Split + win_by_Decision_Unanimous) %>% 
  dplyr::select(-win_by_Decision_Majority, -win_by_Decision_Split, -win_by_Decision_Unanimous, -win_by_TKO_Doctor_Stoppage)

3 K-Means Clustering of Fighters

#Scaling the Data:
fighter_profile_data_reduced <- fighter_profile_data %>% 
  dplyr::select(Referee, #Do not Remove!
         date, #Do not Remove!
         location, #Do not Remove!
         fighter, #Do not Remove!
         Reach_cms, #Please change parameters to be included in clustering here
         Weight_lbs, 
         ends_with("_att"), 
         avg_REV, 
         avg_PASS, 
         avg_KD,
         win_by_Decision,
         win_by_ko_tko,
         win_by_Submission
         ) %>% 
  #what do avg_REV, avg_PASS, avg_KD measure??
  dplyr::select(-starts_with("avg_opp")) %>% 
  dplyr::select(-avg_TOTAL_STR_att, -avg_DISTANCE_att, -avg_SIG_STR_att) %>% 
  na.omit() #filtering all fighters with missing data

fighter_profile_data_scaled <- fighter_profile_data_reduced %>% 
  dplyr::select(-Referee, 
         -date, 
         -location,
         -fighter)
  
fighter_profile_data_scaled <- data.frame(scale(fighter_profile_data_scaled))

fighter_profile_data_scaled %>% 
  ggcorr(method = c("pairwise", "pearson"), layout.exp = 3,label_round=2, label = TRUE,label_size = 2,hjust = 1)

fviz_nbclust(fighter_profile_data_scaled, kmeans, method = "silhouette", k.max = 15) +
  labs(subtitle = "Silhouette method")

## Warning: did not converge in 10 iterations

model_km2 <- kmeans(fighter_profile_data_scaled, centers = 2, nstart = 50, iter.max = 100)
model_km3 <- kmeans(fighter_profile_data_scaled, centers = 3, nstart = 50, iter.max = 100)
model_km4 <- kmeans(fighter_profile_data_scaled, centers = 4, nstart = 50, iter.max = 100)
model_km5 <- kmeans(fighter_profile_data_scaled, centers = 5, nstart = 50, iter.max = 100)

#add clusters to the data frame
fighter_profile_data_scaled_withClusters <- mutate(fighter_profile_data_scaled, cluster = as.factor(model_km2$cluster))

#First let's find the averages of the variables by cluster
center_locations <- fighter_profile_data_scaled_withClusters %>% 
  group_by(cluster) %>% 
  summarize_at(vars(Reach_cms:win_by_Submission),mean)

#Next I use gather to collect information together
xa2 <- gather(center_locations, key = "variable", value = "value", -cluster, factor_key = TRUE)

#Next I use ggplot to visualize centers
knnCenters <- ggplot(xa2, aes(x = variable, y = value)) + 
  geom_line(aes(color = cluster,group = cluster), linetype = "dashed",size=1) +
  geom_point(size=2,shape=4) + 
  geom_hline(yintercept=0) + 
  ggtitle("K-means centers k=2") + 
  labs(fill = "Cluster") + 
  theme(text = element_text(size=10),
        axis.text.x = element_text(angle=45,hjust=1),
        legend.title=element_text(size=5),
        legend.text = element_text(size=5)
  )
knnCenters

fviz_cluster(model_km2, fighter_profile_data_scaled, palette = "Set2", ggtheme = theme_minimal())

#add clusters to the data frame
fighter_profile_data_scaled_withClusters <- mutate(fighter_profile_data_scaled, cluster = as.factor(model_km4$cluster))

#Your code here
#First let's find the averages of the variables by cluster
center_locations <- fighter_profile_data_scaled_withClusters %>% 
  group_by(cluster) %>% 
  summarize_at(vars(Reach_cms:win_by_Submission),mean)

#Next I use gather to collect information together
xa4 <- gather(center_locations, key = "variable", value = "value", -cluster, factor_key = TRUE)

#Next I use ggplot to visualize centers
knnCenters <- ggplot(xa4, aes(x = variable, y = value)) + 
  geom_line(aes(color = cluster,group = cluster), linetype = "dashed",size=1) +
  geom_point(size=2,shape=4) + 
  geom_hline(yintercept=0) + 
  ggtitle("K-means centers k=4") + 
  labs(fill = "Cluster") + 
  theme(text = element_text(size=10),
        axis.text.x = element_text(angle=45,hjust=1),
        legend.title=element_text(size=5),
        legend.text = element_text(size=5)
  )
knnCenters

fviz_cluster(model_km4, fighter_profile_data_scaled, palette = "Set2", ggtheme = theme_minimal())

model_km2$size #3016 3626

## [1] 3016 3626

model_km3$size #2300 2339 2003

## [1] 2300 2339 2003

model_km4$size #2052 1426 1460 1704

## [1] 2052 1426 1460 1704

model_km5$size #1387 1215 1513 1356 1171

## [1] 1387 1215 1513 1356 1171

g2 <- fviz_cluster(model_km2, fighter_profile_data_scaled, palette = "Set2", ggtheme = theme_minimal())
g3 <- fviz_cluster(model_km3, fighter_profile_data_scaled, palette = "Set2", ggtheme = theme_minimal())
g4 <- fviz_cluster(model_km4, fighter_profile_data_scaled, palette = "Set2", ggtheme = theme_minimal())
g5 <- fviz_cluster(model_km5, fighter_profile_data_scaled, palette = "Set2", ggtheme = theme_minimal())
grid.arrange(g2,g3,g4,g5)

##Combining the Data

The following code integrates the cluster in the original fighter-dataframe

fighter_profile_data_reduced_withClusters <- mutate(fighter_profile_data_reduced, cluster = as.factor(model_km4$cluster)) 

temp <- fighter_profile_data_reduced_withClusters %>% 
  dplyr::select(Referee, date, location, fighter, cluster)

fighter_profile_data <- fighter_profile_data %>% 
  left_join(temp, by = c("fighter", "location", "date", "Referee")) %>% 
  mutate(new_fighter = ifelse(is.na(total_time_fought_seconds), 1,0),
         vet_but_no_wins = ifelse(!is.na(total_time_fought_seconds) & wins==0, 1, 0))

This is a sanity check. The function should always output 0:

length(fighter_profile_data$date[is.na(cluster)& fighter_profile_data$new_fighter==0&fighter_profile_data$vet_but_no_wins==0])

## Warning in is.na(cluster): is.na() applied to non-(list or vector) of type
## 'closure'

## [1] 0

total_fight_data <- read_delim("raw_total_fight_data.csv", delim = ";")

## Parsed with column specification:
## cols(
##   .default = col_character(),
##   R_KD = col_double(),
##   B_KD = col_double(),
##   R_SUB_ATT = col_double(),
##   B_SUB_ATT = col_double(),
##   R_PASS = col_double(),
##   B_PASS = col_double(),
##   R_REV = col_double(),
##   B_REV = col_double(),
##   last_round = col_double(),
##   last_round_time = col_time(format = "")
## )

## See spec(...) for full column specifications.

clean_names(total_fight_data)

## # A tibble: 5,144 x 41
##    r_fighter b_fighter  r_kd  b_kd r_sig_str b_sig_str r_sig_str_pct
##    <chr>     <chr>     <dbl> <dbl> <chr>     <chr>     <chr>        
##  1 Henry Ce… Marlon M…     0     0 90 of 171 57 of 119 52%          
##  2 Valentin… Jessica …     1     0 8 of 11   2 of 12   72%          
##  3 Tony Fer… Donald C…     0     0 104 of 2… 68 of 185 52%          
##  4 Jimmie R… Petr Yan      0     2 73 of 192 56 of 189 38%          
##  5 Tai Tuiv… Blagoy I…     0     1 64 of 144 73 of 123 44%          
##  6 Tatiana … Nina Ans…     0     0 75 of 142 48 of 99  52%          
##  7 Aljamain… Pedro Mu…     0     0 174 of 3… 105 of 2… 49%          
##  8 Karolina… Alexa Gr…     0     0 90 of 232 148 of 3… 38%          
##  9 Ricardo … Calvin K…     0     1 12 of 29  22 of 41  41%          
## 10 Yan Xiao… Angela H…     0     0 94 of 249 71 of 144 37%          
## # … with 5,134 more rows, and 34 more variables: b_sig_str_pct <chr>,
## #   r_total_str <chr>, b_total_str <chr>, r_td <chr>, b_td <chr>,
## #   r_td_pct <chr>, b_td_pct <chr>, r_sub_att <dbl>, b_sub_att <dbl>,
## #   r_pass <dbl>, b_pass <dbl>, r_rev <dbl>, b_rev <dbl>, r_head <chr>,
## #   b_head <chr>, r_body <chr>, b_body <chr>, r_leg <chr>, b_leg <chr>,
## #   r_distance <chr>, b_distance <chr>, r_clinch <chr>, b_clinch <chr>,
## #   r_ground <chr>, b_ground <chr>, win_by <chr>, last_round <dbl>,
## #   last_round_time <time>, format <chr>, referee <chr>, date <chr>,
## #   location <chr>, fight_type <chr>, winner <chr>

total_fight_data <- total_fight_data %>% 
  mutate(date = mdy(date)) %>% 
  dplyr::select(-Winner, -Fight_type) %>% 
  left_join(fight_data, by = c("R_fighter", "B_fighter", "Referee", "date", "location")) %>% 
  mutate(women = ifelse(grepl("Women", weight_class), TRUE, FALSE))

total_fight_data_men <- total_fight_data %>% 
  subset(women == FALSE)

total_fight_data_women <- total_fight_data %>% 
  subset(women == TRUE)

total_fight_data_men$R_cluster <- NA
total_fight_data_men$B_cluster <- NA

total_fight_data_men$R_cluster <- fighter_profile_data$cluster[match(paste(total_fight_data_men$location, total_fight_data_men$Referee, total_fight_data_men$date, total_fight_data_men$R_fighter), paste(fighter_profile_data$location, fighter_profile_data$Referee, fighter_profile_data$date, fighter_profile_data$fighter))]

total_fight_data_men$B_cluster <- fighter_profile_data$cluster[match(paste(total_fight_data_men$location, total_fight_data_men$Referee, total_fight_data_men$date, total_fight_data_men$B_fighter), paste(fighter_profile_data$location, fighter_profile_data$Referee, fighter_profile_data$date, fighter_profile_data$fighter))]

sum(is.na(total_fight_data_men$R_cluster)) #1141

## [1] 1141

sum(is.na(total_fight_data_men$B_cluster)) #1877

## [1] 1877

#3018

sum(is.na(fighter_profile_data$cluster)) #3014

## [1] 3014

library(maps)

## 
## Attaching package: 'maps'

## The following object is masked from 'package:cluster':
## 
##     votes.repub

## The following object is masked from 'package:purrr':
## 
##     map

library(maptools)   #mapping: plotting

## Loading required package: sp

## Checking rgeos availability: TRUE

## 
## Attaching package: 'maptools'

## The following object is masked from 'package:Hmisc':
## 
##     label

library(rgeos)      #mapping: simplifying geometry

## rgeos version: 0.5-2, (SVN revision 621)
##  GEOS runtime version: 3.7.2-CAPI-1.11.2 
##  Linking to sp version: 1.3-1 
##  Polygon checking: TRUE

## 
## Attaching package: 'rgeos'

## The following object is masked from 'package:Hmisc':
## 
##     translate

library(rgdal)      #mapping: projecting

## rgdal: version: 1.4-6, (SVN revision 841)
##  Geospatial Data Abstraction Library extensions to R successfully loaded
##  Loaded GDAL runtime: GDAL 2.4.2, released 2019/06/28
##  Path to GDAL shared files: /Library/Frameworks/R.framework/Versions/3.6/Resources/library/rgdal/gdal
##  GDAL binary built with GEOS: FALSE 
##  Loaded PROJ.4 runtime: Rel. 5.2.0, September 15th, 2018, [PJ_VERSION: 520]
##  Path to PROJ.4 shared files: /Library/Frameworks/R.framework/Versions/3.6/Resources/library/rgdal/proj
##  Linking to sp version: 1.3-1

library(spdep)      #Moran's I, Geary's C, poly2nb

## Loading required package: spData

## To access larger datasets in this package, install the spDataLarge
## package with: `install.packages('spDataLarge',
## repos='https://nowosad.github.io/drat/', type='source')`

## Loading required package: sf

## Linking to GEOS 3.7.2, GDAL 2.4.2, PROJ 5.2.0

## 
## Attaching package: 'spdep'

## The following object is masked from 'package:fmsb':
## 
##     geary.test

library(rnaturalearth)
library(ggmap)

## Google's Terms of Service: https://cloud.google.com/maps-platform/terms/.

## Please cite ggmap if you use it! See citation("ggmap") for details.

library(opencage)
library(sf)
library(gganimate)
library(hrbrthemes)

map_data <- total_fight_data_men %>% 
  group_by(location) %>% 
  summarise(count = n(),
         first_fight = min(date))

latitudes <- numeric(0)
longitudes <- numeric(0)
locations <- map_data$location

for (i in locations){
   temp = opencage_forward(i, key = OPENCAGE_KEY)
   lat = median(temp$results$geometry.lat)
   lng = median(temp$results$geometry.lng)
   latitudes[i] = lat
   longitudes[i] = lng
}

map_data <- map_data %>% 
  mutate(lat = latitudes,
         lng = longitudes)

world <- ne_countries(scale = "medium", returnclass = "sf") %>% 
  filter(name != "Antarctica")

plt <- ggplot(world) +
  geom_sf() +
  geom_point(data=map_data,
             aes(x=lng,
                 y=lat,
                 size=count,
                 color=first_fight,
                 group = seq_along(first_fight)))+
  labs(title = "The History of UFC: How fights have expanded globally over time")+
  theme(text=element_text(size = 13, family="Tahoma"))+
  transition_reveal(first_fight) +
  shadow_mark(alpha = 0.3, size = 0.5)
  
plt

#Plot a distribution curve of how long each fight lasted, split by round format
# add last round time, total fight time
total_fight_data_men <- total_fight_data_men %>% 
  mutate(last_round_time = as.character(last_round_time),
         last_round_time = gsub('.{3}$', '', last_round_time),
         last_round_time = ms(last_round_time),
         total_fight_time = (last_round-1)*ms("05:00")+last_round_time,
         total_fight_time_sec = seconds(total_fight_time))

# total time = (last_round - 1) * 5 + (last_round time)
density_data <- total_fight_data_men %>% 
  mutate(no_of_rounds = as.factor(no_of_rounds))

vertical.lines <- c(300,600,900,1200,1500)
density <- ggplot(density_data, aes(x = total_fight_time_sec, color = no_of_rounds, fill = no_of_rounds)) + geom_density(alpha = 0.9) + geom_vline(xintercept = vertical.lines, linetype = "dotted", color = "orange", size = 0.5) +
  labs(title = "How long do the fights of different formats last?", x = "Fight Time (sec)", y = "Density", fill = "Round Format", color = "Round Format")+
  theme_light()+
  scale_fill_brewer(palette = "Accent")+
  scale_color_brewer(palette = "Accent")+
  theme(text=element_text(family="Tahoma"))
density

We now have 2 table:

total_fight_data_men containing information on the fights. fighter_profile_data containing information on the fighters (incl. the outcome of the clustering)

The two tables can be linked using the Referee, location, date and fighter column(s).

ToDO:

*Parsing of columns in the total_fight_data table (eg. getting rid of “of”s in some columns)

3.1 Visualizations

# GGRadar Plot

# Data Processing before making a plot
cluster_radar <- center_locations %>% 
  mutate_each(funs(rescale), -cluster) %>% 
  rename("Arm Reach" = Reach_cms, "Weight(lbs)" = Weight_lbs, "Body" = avg_BODY_att, "Clinch" = avg_CLINCH_att, "Ground" = avg_GROUND_att,
         "Head" = avg_HEAD_att, "Leg" = avg_LEG_att, "Sub" = avg_SUB_ATT, "TD" = avg_TD_att, "REV" = avg_REV, "PASS" = avg_PASS, 
         "K/D" = avg_KD, "Decision" = win_by_Decision, "TKO or KO" = win_by_ko_tko, "Submission" = win_by_Submission) %>%
  mutate(cluster = as.numeric(cluster))

## Warning: funs() is soft deprecated as of dplyr 0.8.0
## Please use a list of either functions or lambdas: 
## 
##   # Simple named list: 
##   list(mean = mean, median = median)
## 
##   # Auto named with `tibble::lst()`: 
##   tibble::lst(mean, median)
## 
##   # Using lambdas
##   list(~ mean(., trim = .2), ~ median(., na.rm = TRUE))
## This warning is displayed once per session.

# rename clusters to our self-defined clusters
cluster_radar[1, 1] <- "Heavy Fighters"
cluster_radar[2, 1] <- "Submission Artists"
cluster_radar[3, 1] <- "Wrestlers"
cluster_radar[4, 1] <- "All-Rounders"

glimpse(cluster_radar)

## Observations: 4
## Variables: 16
## $ cluster       <chr> "Heavy Fighters", "Submission Artists", "Wrestlers", "A…
## $ `Arm Reach`   <dbl> 1.0000000, 0.2798764, 0.5721881, 0.0000000
## $ `Weight(lbs)` <dbl> 1.0000000, 0.2397387, 0.3357986, 0.0000000
## $ Body          <dbl> 0.1255582, 0.2282015, 0.0000000, 1.0000000
## $ Clinch        <dbl> 0.3605794, 0.3431903, 0.0000000, 1.0000000
## $ Ground        <dbl> 0.12062397, 1.00000000, 0.00000000, 0.03476932
## $ Head          <dbl> 0.1756776, 0.3854319, 0.0000000, 1.0000000
## $ Leg           <dbl> 0.1634816, 0.1496618, 0.0000000, 1.0000000
## $ Sub           <dbl> 0.00000000, 0.51160938, 1.00000000, 0.06916931
## $ TD            <dbl> 0.0000000, 1.0000000, 0.2775629, 0.3232438
## $ REV           <dbl> 0.0000000, 1.0000000, 0.6626501, 0.2257382
## $ PASS          <dbl> 0.00000000, 1.00000000, 0.33575056, 0.05968277
## $ `K/D`         <dbl> 1.0000000, 0.0128156, 0.0000000, 0.3253070
## $ Decision      <dbl> 0.02575786, 0.96508211, 0.00000000, 1.00000000
## $ `TKO or KO`   <dbl> 1.0000000, 0.1166366, 0.0000000, 0.1596701
## $ Submission    <dbl> 0.00000000, 0.09411239, 1.00000000, 0.02505515

# ggradar plot
ggradar_plot <- cluster_radar %>% 
  ggradar(font.radar = "Tahoma",
        background.circle.colour = "#D7D6D1",
        background.circle.transparency = 0.1,
        plot.legend = if (nrow(cluster_radar) > 1) TRUE else FALSE,
        legend.title = "Fighter Type",
        legend.text.size = 10,
        plot.title = "4 Clusters of UFC",
        group.point.size = 3,
        base.size = 14,
        plot.extent.x.sf = 1,
        plot.extent.y.sf = 1.2,
        grid.label.size = 5,
        axis.label.size = 3.5) #It would be nice if we can name the types of the fighters P!

plot(ggradar_plot)

# Individual Plots
ggRadar_plot <- ggRadar(data = cluster_radar, aes(color = cluster),legend.position = "none", interactive = FALSE, size = 1, rescale = FALSE, use.label = TRUE)+
  scale_y_discrete(breaks = NULL)+
  ggtitle("Stat by Each UFC Fighter Types")+
  theme(axis.text.x = element_text(size = 5))+
  scale_color_brewer(palette = "Set2")+
  scale_fill_brewer(palette = "Set2")+
  theme(text=element_text(size=13, family="Tahoma"))+
  facet_grid(~cluster)+
  theme(axis.text.x = element_text(vjust = 50, angle = -90-360/15*(1:15)))

plot(ggRadar_plot)

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (0 <= angle & angle < 90) {: the condition has length > 1 and only
## the first element will be used

## Warning in if (90 <= angle & angle < 180) {: the condition has length > 1 and
## only the first element will be used

## Warning in if (180 <= angle & angle < 270) {: the condition has length > 1 and
## only the first element will be used

fighter_data <- fighter_profile_data %>%
  dplyr::select(-Referee:-fighter) %>%
  na.omit()

fighter_data$Weight_lbs = as.factor(fighter_data$Weight_lbs)

fighter_data$Weight_lbs <- fct_collapse(fighter_data$Weight_lbs,
  '125' = ("125"),
  '135' = ("135"),
  '145' = ("145"),
  '155' = ("155"),
  '170' = c("168", "170"),
  '185' = ("185"),
  '205' = ("205"),
  heavy = c("225", "230", "231", "234", "235", "238", "240", "242", "243", "245", "247", "249", "250", "253", "255", "258",
             "260", "263", "264", "265"))


f6 <- fighter_data %>%
      group_by(cluster) %>%
    dplyr::summarise(number=dplyr::n())

f5 <- fighter_data %>%
  group_by(cluster, Weight_lbs) %>%
  dplyr::summarise(number=dplyr::n()) 


for (i in 1:8){
  f5$number[i] = f5$number[i]/2037
}


for (i in 9:16){
  f5$number[i] = f5$number[i]/1406
}


for (i in 17:24){
  f5$number[i] = f5$number[i]/1443
}


for (i in 25:32){
  f5$number[i] = f5$number[i]/1664
}


ggplot(f5, aes(x = cluster, y =number)) + 
geom_col(aes(fill = Weight_lbs))+
  theme_bw()+
  labs(title="Clusters and Weight Classes ",
       subtitle = "What are the proportion of different weight classes in each cluster?", 
       x = "Cluster Number",
       y = "Proportion of fighters")+
  theme(text=element_text(size=13, family="Tahoma"))+
  scale_fill_brewer(palette = "Greens")

  NULL

## NULL

f5$Weight_lbs = as.factor(f5$Weight_lbs)


f5$Weight_lbs <- fct_collapse(f5$Weight_lbs,
  '125' = ("125"),
  '135' = ("135"),
  '145' = ("145"),
  '155' = ("155"),
  '170' = c("168", "170"),
  '185' = ("185"),
  '205' = ("205"),
  heavy = c("225", "230", "231", "234", "235", "238", "240", "242", "243", "245", "247", "249", "250", "253", "255", "258",
             "260", "263", "264", "265"))

## Warning: Unknown levels in `f`: 168, 225, 230, 231, 234, 235, 238, 240, 242,
## 243, 245, 247, 249, 250, 253, 255, 258, 260, 263, 264, 265

f <- fighter_data %>%
  group_by(Weight_lbs) %>%
  dplyr::summarise(number=dplyr::n(),
              win_by_ko_tko= sum(win_by_ko_tko),
            win_by_Submission = sum(win_by_Submission),
            win_by_Decision = sum(win_by_Decision)) 

f$Weight_lbs = as.factor(f$Weight_lbs)


f$Weight_lbs <- fct_collapse(f$Weight_lbs,
  '125' = ("125"),
  '135' = ("135"),
  '145' = ("145"),
  '155' = ("155"),
  '170' = c("168", "170"),
  '185' = ("185"),
  '205' = ("205"),
  heavy = c("225", "230", "231", "234", "235", "238", "240", "242", "243", "245", "247", "249", "250", "253", "255", "258",
             "260", "263", "264", "265"))

## Warning: Unknown levels in `f`: 168, 225, 230, 231, 234, 235, 238, 240, 242,
## 243, 245, 247, 249, 250, 253, 255, 258, 260, 263, 264, 265

f1 <- f %>%
  group_by(Weight_lbs) %>%
  dplyr::summarise(number = sum(number),
              win_by_ko_tko= sum(win_by_ko_tko),
            win_by_Submission = sum(win_by_Submission),
            win_by_Decision = sum(win_by_Decision)) 
  

for (i in 1:nrow(f1)){
  f1$win_by_ko_tko[i] = f1$win_by_ko_tko[i]/f1$number[i]*100
  f1$win_by_Submission[i] = f1$win_by_Submission[i]/f1$number[i]*100
  f1$win_by_Decision[i] = f1$win_by_Decision[i]/f1$number[i]*100
}

f1_long <- gather(f1, win_type, percentage, win_by_ko_tko:win_by_Decision, factor_key=TRUE)

# Rename the values in the dataframe to make the legend more clear
levels(f1_long$win_type)[levels(f1_long$win_type) == "win_by_ko_tko"] <- "Win by KO/TKO"
levels(f1_long$win_type)[levels(f1_long$win_type) == "win_by_Submission"] <- "Win by Submission"
levels(f1_long$win_type)[levels(f1_long$win_type) == "win_by_Decision"] <- "Win by Decision"


ggplot(f1_long, aes(x = Weight_lbs, y =percentage)) + 
geom_col(aes(fill = win_type))+
  theme_bw()+
  labs(title="Win Type",
       subtitle = "Variation of win types between weigh classes", 
       x = "Weight Class",
       y = "Percentage of Wins",
       fill = "Win Type")+
  scale_fill_brewer(palette = "Paired")+
  theme(text=element_text(family="Tahoma"))+
  NULL

#Calculating summery data for the cluster matchups
cluster_perfA <- total_fight_data_men %>% 
  group_by(R_cluster, B_cluster, Winner) %>% 
  summarise(fights = n()) 

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_perfB <- total_fight_data_men %>% 
  group_by(B_cluster, R_cluster, Winner) %>% 
  summarise(fights = n())

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_perfA <- cluster_perfA %>% 
  spread(key = Winner, value = fights) 

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_perfB <- cluster_perfB %>% 
  spread(key = Winner, value = fights)

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

library(tidyr)
cluster_perfA$Draw <- tidyr::replace_na(cluster_perfA$Draw, 0) 
cluster_perfB$Draw <- tidyr::replace_na(cluster_perfB$Draw, 0)

cluster_perfA <- cluster_perfA %>% 
  mutate(fights = Blue + Draw + Red) %>% 
  rename(fighterCluster=R_cluster,
         oppCluster=B_cluster,
         Wins=Red,
         Losses=Blue)

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_perfB <- cluster_perfB %>% 
  mutate(fights = Blue + Draw + Red) %>% 
  rename(fighterCluster=B_cluster,
         oppCluster=R_cluster,
         Wins=Blue,
         Losses=Red)

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_perf_incl_draws <- cluster_perfA %>% 
  left_join(cluster_perfB, by = c("fighterCluster", "oppCluster")) %>% 
  summarise(fights = (fights.x + fights.y),
            wins = (Wins.x + Wins.y),
            losses = (Losses.x + Losses.y),
            draws = (Draw.x + Draw.y)
            ) %>% 
  mutate(win_per = wins/fights,
         draw_per = draws/fights)

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `oppCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_perf_excl_draws <- cluster_perfA %>% 
  left_join(cluster_perfB, by = c("fighterCluster", "oppCluster")) %>% 
  summarise(fights = (fights.x + fights.y - Draw.x - Draw.y),
            wins = (Wins.x + Wins.y),
            losses = (Losses.x + Losses.y)
            ) %>% 
  mutate(win_per = wins/fights)

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `oppCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

#Create Heatmap for Wins
dfWins <- cluster_perf_excl_draws %>% 
  dplyr::select(-fights, -wins, -losses) %>% 
  spread(key = oppCluster, value = win_per) 

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

#dfWins[1, 3:6] = NA
#dfWins[2, 4:6] = NA
#dfWins[3, 5:6] = NA
#dfWins[4, 6] = NA

dfWins <- dfWins %>% 
  gather(2:6, key = oppCluster, value = win_per)

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

dfWins$fighterCluster <- ifelse(is.na(dfWins$fighterCluster), 'newFighter', dfWins$fighterCluster)
dfWins$oppCluster <- ifelse(dfWins$oppCluster=="<NA>", 'newFighter', dfWins$oppCluster)

dfWins <- dfWins %>% 
  ungroup() %>% 
  mutate(fighterCluster = factor(fighterCluster, levels = c(1,2,3,4,"newFighter"))) %>% 
  group_by(fighterCluster) %>% 
  mutate(oppCluster = factor(oppCluster, levels = c(1,2,3,4,"newFighter")))
 
library(forcats)
ggplot(dfWins, aes(x = oppCluster, y = fct_rev(fighterCluster), fill = win_per)) +
  geom_tile()+
  geom_text(aes(label=round(win_per, digits=2)))+
  theme(text=element_text(size = 13, family="Tahoma"))+
  labs(
    title = "Heatmap for Probability of Win",
    x = "Opponent Cluster",
    y = "Fighter Cluster",
    fill = "Probability of Win (%)"
  )+
  NULL

#Create Heatmap for Draws
dfDraws <- cluster_perf_incl_draws %>% 
  dplyr::select(-fights, -wins, -losses, -draws, -win_per) %>% 
  spread(key = oppCluster, value = draw_per) 

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

#dfDraws[1, 3:6] = NA
#dfDraws[2, 4:6] = NA
#dfDraws[3, 5:6] = NA
#dfDraws[4, 6] = NA

dfDraws <- dfDraws %>% 
  gather(2:6, key = oppCluster, value = draw_per)

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

dfDraws$fighterCluster <- ifelse(is.na(dfDraws$fighterCluster), 'newFighter', dfDraws$fighterCluster)
dfDraws$oppCluster <- ifelse(dfDraws$oppCluster=="<NA>", 'newFighter', dfDraws$oppCluster)

dfDraws <- dfDraws %>% 
  ungroup() %>% 
  mutate(fighterCluster = factor(fighterCluster, levels = c(1,2,3,4,"newFighter"))) %>% 
  group_by(fighterCluster) %>% 
  mutate(oppCluster = factor(oppCluster, levels = c(1,2,3,4,"newFighter")))
 
ggplot(dfDraws, aes(x = oppCluster, y = fct_rev(fighterCluster), fill = draw_per)) +
  geom_tile()+
  geom_text(aes(label=round(draw_per, digits=3)))+
  theme(text=element_text(size = 13, family="Tahoma"))+
  labs(
    title = "Heatmap for Probability of Draws",
    x = "Opponent Cluster",
    y = "Fighter Cluster",
    fill = "Probability of Draws (%)"
  )+
  NULL

#Calculating summery data for average fight time
cluster_timeA <- total_fight_data_men %>% 
  filter(no_of_rounds==3) %>% #filter for modern standart duration only
  group_by(R_cluster, B_cluster) %>% 
  summarise(avg_length = mean(total_fight_time_sec),
            fights = n()) %>% 
  rename(fighterCluster=R_cluster,
         oppCluster=B_cluster)

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_timeB <- total_fight_data_men %>% 
  filter(no_of_rounds==3) %>% #filter for modern standart duration only
  group_by(B_cluster, R_cluster) %>% 
  summarise(avg_length = mean(total_fight_time_sec),
            fights = n()) %>% 
  rename(fighterCluster=B_cluster,
         oppCluster=R_cluster)

## Warning: Factor `B_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `R_cluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

cluster_time <- cluster_timeA %>% 
  left_join(cluster_timeB, by = c("fighterCluster", "oppCluster")) %>% 
  mutate(avg_length = ((avg_length.x * fights.x) + (avg_length.y * fights.y)) / (fights.x + fights.y),
         fights = fights.x + fights.y) %>% 
  dplyr::select(-avg_length.x, -fights.x, -avg_length.y, -fights.y)

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

dfTime <- cluster_time %>% 
  dplyr::select(-fights) %>% 
  spread(key = oppCluster, value = avg_length) 

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

#dfTime[1, 3:6] = NA
#dfTime[2, 4:6] = NA
#dfTime[3, 5:6] = NA
#dfTime[4, 6] = NA

dfTime <- dfTime %>% 
  gather(2:6, key = oppCluster, value = avg_length)

## Warning: Factor `fighterCluster` contains implicit NA, consider using
## `forcats::fct_explicit_na`

dfTime$fighterCluster <- ifelse(is.na(dfTime$fighterCluster), 'newFighter', dfTime$fighterCluster)
dfTime$oppCluster <- ifelse(dfTime$oppCluster=="<NA>", 'newFighter', dfTime$oppCluster)

dfTime <- dfTime %>% 
  ungroup() %>% 
  mutate(fighterCluster = factor(fighterCluster, levels = c(1,2,3,4,"newFighter"))) %>% 
  group_by(fighterCluster) %>% 
  mutate(oppCluster = factor(oppCluster, levels = c(1,2,3,4,"newFighter"))) 

dfTime$avg_length <- seconds(round(dfTime$avg_length))
#dfTime$avg_length <- ms(dfTime$avg_length)
 
ggplot(dfTime, aes(x = oppCluster, y = fct_rev(fighterCluster), fill = avg_length)) +
  geom_tile()+
  geom_text(aes(label=round(avg_length, digits=3)))+
  theme(text=element_text(size = 13, family="Tahoma"))+
  labs(
    title = "Heatmap of Average Fight Time",
    x = "Opponent Cluster",
    y = "Fighter Cluster",
    fill = "Avergae Fight Time (sec)"
  )+
  NULL